Banks Are Now Handing Out Loans to People They’d Normally Shun

Before Jeff Stewart could credibly advance the notion that your Facebook friend network could—or even should—help you get a loan, he had to prove the system worked and was reliable. It took a long time. I first met him in 2012, a year after he’d launched Lenddo, pitching it as a new way for people just entering the middle class to qualify for and receive loans. His startup planned to use Facebook data to assess credit risk, and to make loans to well-qualified candidates, starting in the Philippines.

I thought it sounded like an interesting idea, but hey, Facebook was still mostly party pictures and a lot of flirting. Hell, my mom had just joined. Trying to imagine the platform would ever be a tool for judging borrowers’ creditworthiness seemed like a stretch.

Three years later, Lenddo has built a thriving business lending money to members of the emerging middle class around the globe. More than 250,000 people have joined Lenddo in its first market, the Philippines. The company, which WIRED wrote about last year, has also launched in Colombia, Mexico, Indonesia and parts of Africa.

In the process, Stewart has tweaked his algorithm, which pulls data from social media sites, so that it has become a reliable predictor of who is likely to pay back loans. “We had to test out all the parts of our thesis—including making loans to people we knew wouldn’t pay us back—to make sure it worked,” Stewart told me when we spoke last week. “By last summer, we finally had enough data to show that the score worked.” The startup’s default rates, according to Stewart, were in the low single digits, mirroring the average default rates in the microfinance industry.

Now that Stewart can prove that his scoring system is dependable, his company has stopped lending. Starting this month, it will instead sell its algorithm, the Lenddo score, to financial institutions, which will use it to make loans. In other words, Lenddo hopes to become, for the Filipino call center worker, what Experian is for me here in the US: the provider of the definitive measure of financial trustworthiness.

From the outset, Lenddo has believed that the digital exhaust of loan candidates offers better evidence that they can repay a loan than what’s provided by a traditional credit score—especially for people just entering the middle class, a group Stewart estimates to include 1.2 billion people worldwide. Their behavior on these networks, as well as the ties they have to others, provides a rich source of data about their character. The Hong Kong-based company has $14 million in capital from Accel Partners and other investors, and it has struck deals with a number of banks. Its customers are institutions like BanKO, the mobile-phone based savings bank that is a joint venture between the Bank of the Philippines Islands, Globe Telecom and Ayala Corporation.

As Internet companies, telecoms and banks all attempt to bring banking to this new group of borrowers, however, the competition is heating up. “There are millions of great credit-worthy customers out there, but today, most of their financial activity is informal and cash based,” says James Moed, a London-based financial services consultant helping companies design new products for the emerging middle class. “Whoever can efficiently translate that behavior into a reliable scoring will have borrowers lining up.” A number of startups like Cignifi are attempting to judge loan risk for these borrowers by analyzing their cell phone activity. Others determine how qualified a loan candidate might be through online personality and trustworthiness tests. Then there’s Affirm, the startup Max Levchin founded that tries to improve upon the commonly used FICO score by incorporating social data among other metrics to judge creditworthiness for consumers (particularly millennials).

With three years’ worth of data, Lenddo has a big head start in the international market. It has nearly a million members on its network who give Lenddo access to their social activity on Twitter, Facebook, LinkedIn, Google, Yahoo, and Hotmail. Lenddo’s software mines that data to see who members communicate with, and how often, and then transforms that information into a score. Says Stewart, “We don’t share any of the data, only a score. Full stop.”

In some cases, Lenddo also asks members to select a group of “trusted friends” as references, drawing from a proven microfinance strategy. If they’re unable to pay back the loan, their references will be deemed less creditworthy.

To become a useful lending tool, however, the Lenddo score must be a recognizable consumer brand. Loan recipients must know to sign up for one, either because it will help them get a loan with their local bank, or because their upstanding social behavior will help their friends get a loan. In its earliest market, that has already happened. Says Stewart, “If you are Filipino, you have among your closest 150 friends those who use Lenddo.”

Now, Lenddo just needs to make that a reality in 195 more countries.